CA2524224C - Method for transmitting data by means of a carrier current - Google Patents

Abstract

The invention relates to a method for transmitting data over a carrier current using a frequency band. According to said method: the frequency band is divided into N sub-bands, N being a whole number higher than, or equal to, two; an OFDM (Orthogonal Frequency Division Multiplexing) technique is carried out on each of said sub-bands; and calculations are carried out based on each sub-band. The inventive method is characterised in that the calculations based on each sub-band are independent from the calculations based on the other sub-bands, and said sub-bands are dynamically activated and allocated. The invention also relates to modulation and demodulation devices for carrying out said method.

Description

WO 2004/10039

2 PCT / FR2004 / 001057 METHOD FOR DATA TRANSMISSION BY POWER CURRENT
The present invention relates to the field of powerline data transmission.
The present invention relates more particularly to a method of data transmission and to a modem implementing the OFDM technique (Orthogonal Frequency Division Multiplexing).
The prior art already knows, by the patent application EP 1 011 235 (Nortel), a communication system using OFDM on the carrier current. Management of the noise is added by "clipping" the incoming signal.
The prior art also knows, by the request of European patent EP 1 014 640 (Nortel), a system of communication using the OFDM technique. Solutions are made for the synchronization between the base and the slaves. Special symbols are added and a guard longer is used.
It is proposed in the PCT patent application WO
01/95518 (Conexant), a method and a device for dual-band modulation in network systems of carrier current. This invention implements the technique OFDM and allows you to switch from one band to another static.
It is also proposed in the patent application EP 1 018 826 (PolyTrax), a method for Powerline data transmission using the OFDM technique. In the invention presented in this document, we improve the signal either by implementing a window not rectangular on the time signal (one carrier out of two is lost), or by implementing a filter whose coefficients are given.

The prior art also knows, by the patent US 5,610,908 (BBC), a system for reduce the cc peak. to mean ratio> a.
The prior art also knows, by the patent US 6,473,453 (Secretary of State for Defense British), a high-level communication system frequencies implementing a transmission solution of eight multiplexed channels. The eight channels contain signals spread in frequencies. This implementation is interesting for the suppression of interference related to narrow bands.
The prior art also knows, by the patent US 5,684,450 (Norweb), a transmission network using the OFDM technique on the current carrier consisting of implementing filters and a part analogue for the injection of a signal greater than 1 MHz.
A termination unit for calculating the good impedance is provided.
The prior art also knows, by the patent US 6,282,405 (Norweb), a hybrid network of distribution of electricity and telecommunications putting in a very general way the OFDM technique.
The prior art also knows, by the patent US 6,456,657 (Bell Canada), a system of benches filters implementing the OFDM technique. This system allows to break down a broadband system, to multiplex each of the subbands on the channel and rebuild the signal in reception.
The prior art also knows, by the request of PCT Patent WO 02/51089 (Conexant), a system synchronizing on periodic noises using OFDM
on the carrier current. The purpose of this invention is to send the data symbols between two noise peaks.

3 The prior art also knows, by the patent US 6,373,377, a feed with a coupling for data transmission.
Also known in the state of the art, the US Patent 6,249,213 (INTEL) ~ e .1 ~ 'eth ~ df ~ r transmitting informati ~ n ~ uer an alternating current power Thins thr ~ u ~ ha ~ aluralit ~ ~ ffre ~ uency ~ rth ~~~ nal aubchannels aa. An embodiment of the invention described in this US patent is a method for the transmission d ° information through a power line alternative. A frequency channel to transmit the information is selected and divided into a plurality of orthogonal subchannels of frequency. Each subchannel orthogonal frequency is tested to determine a value of a characteristic transmission, which in turn is used to determine a transmission bit density for the orthogonal frequency subchannel. Information are divided into a plurality of data sub-blocks.
Each sub-block of data, corresponds to one of the channels orthogonal frequency, and the size of each sub-block of data is determined based on the bit density corresponding transmission. The data of each sub-block are modulated for transmission through the sub-corresponding frequency orthogonal channel. Each sub-data block is transmitted approximately simultaneously through the orthogonal frequency subchannel corresponding.
This US patent US 6,249,213 to Intel relates well the transmission of data on a power line.
However, he does not mention the steps dynamic activation and allocation requirements of bands. These steps are very important in the process according to the present invention. Moreover, the fact of realizing calculations on each of the sub-bands totally

4 independent of the calculations made on the other sub-bands does not appear in the prior art either.
The present invention intends to remedy disadvantages of the prior art by cutting the strip of frequency in N sub-bands, where N is a higher integer or equal to two, by performing calculations on each of the sub-bands that are independent of the calculations made on the other subbands and by activating the subbands dynamic by software means. In addition, synchronization OFDM symbols is made simple and inexpensive by the use of OFDM symbols smaller than 512 carriers.
For this purpose, the present invention relates, in its the most general meaning, a method of transmitting powerline data using a frequency band consists in .
~ cutting said frequency band into N subbands, N being an integer greater than or equal to two;
~ implement an OFDM technique (Orthogonal Frequency Division Multiplexing) on each of these sub-bands;
~ perform calculation operations on each of the sub-bands;
characterized in that ~ the calculations made on each of the subbands are independent of the calculations made for the other sub-bands;
~ the said subbands are activated and allocated from dynamic way.
According to a particular mode of implementation, the number N of sub-bands is equal to 7.

According to one particular embodiment, the OFDM symbols have a size equal to 256 carriers.
According to an advantageous variant, the subbands are dynamically activated and allocated by means of

5 software.
According to another variant, the subbands are dynamically activated and allocated by a means equipment.
The invention also relates to equipment for modulation and demodulation for the implementation of the process comprising filtering means, means for calculating FFT, Analog / Digital conversion, amplification means and activation means and subband allocation.
The invention will be better understood by means of the description, given below for purely explanatory purposes, of an embodiment of the invention, with reference to annexed figures.
~ Figure 1 represents the general principles of the process according to the invention;
~ Figure 2 illustrates the analog part of reception;
~ Figure 3 illustrates a description of the bands;
~ Figure 4 represents the behavior of the analysis filter;
~ Figure 5 shows the synchronization flow on a sub-band;
~ Figure 6 illustrates the correlation principle of frame symbol;
~ Figure 7 shows the behavior of the block of synchronization;
~ Figure 8 illustrates the principle of equalization and correction of the frequency error;

6 ~ Figure 9 represents the demodulation flow on a sub-band;
~ Figure 10 illustrates the configuration of the modulation on a subband; and ~ Figure 11 shows the analog part of transmission.
The method according to the present invention is based on the use of a frequency band for powerline data transmission. a non example limiting is to use the 1.6 MHz band at 30 MHz.
The principle of the method according to the invention is to cut the band in several sub-bands (two or more), independent and enabled dynamically, using, on each of her, an OFDM technique.
Powerline transmission uses advanced signal processing techniques by modulating authorized frequencies (eg from 1.6 to 30 MHz) on power supply lines by capacitive coupling or inductive. Coupling makes it possible to ensure that high frequency signals (above 50 Hz) and low voltage are transposed on the lines power supply and that, on the other hand, the signals high frequency present on the electrical cables be recovered in the system.
For reception, the modulated signal is retrieved from the power line, it is filtered roughly (1, 6 to 30 MHz) and then amplified. It is then scanned to be analyzed according to the sub-band it occupies. Each sub-band is treated in the same way one after the other.
The signal is synchronized, multiplied to remove the Frequency errors, then passed into the domain frequency. There, we draw some information on

7 the evolution of the canal to better correct the deformations due to the channel and then demodulates. Finally, the control block retrieves the information to transmit it to the layers higher.
For transmission, the control block sends the subband data after subbands. They are modulated then spent in the time domain. The information is then translated on the sub-band that they are supposed to occupy by the synthesis filter. There, the signal digital is converted into an analog signal, filtered for eliminate the components beyond 1.6-30 MHz, then amplified. Finally, the signal is transmitted on the wires electrical coupling.
The transmit and receive channels are pretty different and are found only at the level of control and coupling, as shown in FIG.
here is the detailed description.
Reception - Analog part Reception filter the high frequency signals to keep only the band useful (1.6 to 30 MHz in our exemplary embodiment of the invention) from a series capacitor for low frequencies and a low-pass filter. here is an example of filter characteristics.
- Order elliptical filter 8 - Bandwidth 0-30 MHz - Stop band at -50dB
- Ripple in the bandwidth <1dB
- Bandwidth transition -> Stop band <2MHz It then amplifies the signal with variable gain to adjust the signal dynamics.
Finally, an analog-to-digital converter digitize the signal.
For example - 14 bits and 64 MHz

8 Figure 2 depicts the analog reception part.
Fgltrage L ~. first stage of digital signal processing is the filtering that is central to the innovation of the present invention. In this embodiment of the invention, we will consider seven subbands. I1 is heard that this example can not be restrictive. Filtering allows to recover on 7 different channels, the signals contained in the following 7 sub-bands (TBC).
~ Subband 1. -2 -> 10 MHz ~ Subband 2. 2 -> 12 MHz ~ Subband 3. 6 -> 18 MHz ~ Subband 4. 10 -> 22 MHz ~ Subband 5. 14 -> 26 MHz ~ Subband 6. 18 -> 30 MHz ~ Subband 7. 22 -> 34 MHz Filtering is actually closely related to structure OFDM of the signal. We therefore allow for an overlap of different subbands assuming that the carriers present in the neighboring sub-bands will not be used. This trick reduces the complexity of filters while maintaining high spectral efficiency.
Here are the useful frequencies in each of the subbands.
~ Subband 1. 2 -> 6 MHz ~ Subband 2. 6 -> 10 MHz ~ Subband 3. 10 -> 14 MHz ~ Subband 4. 14 -> 18 MHz ~ Subband 5. 18 -> 22 MHz ~ Subband 6. 22 -> 26 MHz Subband 7. 26 -> 30 MHz Figure 3 shows the different bands.

9 Filtering is done using a filter bank analysis. This structure makes it possible to isolate the signals of each of the 7 sub-bands and to transpose them into based. Eight samples of the global band allow to have a sample per subband. They are multiplexed in time out of the filter.
Figure 4 illustrates the behavior of the filter bank analysis.
Synchronisati ~ n Synchronization between a transmitter and a receiver is done on each band independently. The samples are therefore processed and stored one after the others according to the sub-band concerned. The synchronization is based on the detection of a symbol auto-correlated and a guard interval. She allows to place the window of the Fourier transform to pass in the frequency domain.
Figure 5 shows the synchronization flow on a sub-band.
The synchronization symbol is sent at the beginning of frame. I1 contains known information repeated in the time. The correlation on half of the symbol allows locate this symbol among others. Then, at the end of symbol, we use the guard interval which is the copying from the beginning of the symbol. The correlation over the length of the guard makes it possible to precisely find the window of Fourier transform.
Figure 6 illustrates the correlation principle of frame symbol.
Once an entire symbol is received and detected, the synchronization block sends it to the Fourier transform after a complex multiplication. This allows a correction of the temporal signal from the calculations made further downstream in the reception. At the beginning of the frame, the correction is inactive.
from there, the samples of 5 different subbands are now dissociated. Money-tapes are processed one after the other, 256 samples by 256.
Figure 7 illustrates the behavior of the block of synchronization.
Fourier Transform In a preferred embodiment, the OFDM signal in each of the subbands is composed of 256 carriers.
Even though only 128 carriers are used to respect the structure of the subbands, it is necessary to realize a Fourier transform of 256 points. For 256 temporal samples received in a subband, she calculates 256 frequency samples corresponding to 256 carriers.
The structure used is relatively conventional, based on the butterfly principle. For 256 points including 128 non-zero, this structure has been particularly optimized in terms of surface and speed. It allows the calculation with a latency of 256 cycles approximately.
Frequency correction The electric current, like most media of communications, creates errors on the frequencies that one uses. There may be errors in amplitude and phase, these two parameters drifting in time. For their correction, the solution of the present invention is on the use of two tools. a reference symbol fully known containing all the useful carriers and sent at regular intervals, reference carriers (drivers) in each data symbol. Their location is known to the transmitter and the receiver.
On a reference symbol, the correction is estimated phase and amplitude for all carriers.
The following symbols apply the values found above and we look at the pilot carriers.
For these deus-1 ~~ one calculates the angle of rotation and one in deduce the angle of rotation for all other carriers by linear regression. The angle of rotation of the carrier of the middle is called Delta and this angle is applied on the next symbol in the time domain. Figure 5 represents the synchronization stream on a subband and Figure 8 illustrates the principle of equalization and correction of the frequency error.
demodulation The carriers being finally corrected, it is possible to demodulate them. For this, the transmitter and the receiver are agreed to use modulations specific for each of the 16 groups of 8 carriers per subband. On each of these groups, one can use a modulation adapted to the noise level.
0 - no modulation 1 - phase modulation (BPSK) 2 - quadrature phase modulation (QPSK) 4 - amplitude and quadrature modulation 16 (~ AM16) 6 - amplitude and quadrature modulation 64 (QP ~ NI6 4) 8 - amplitude modulation and quadrature 256 (~ ANI256) Reference symbols and carriers drivers are not demodulated. Moreover, on each of certain carriers are forbidden because used by other priority services. They are not not demodulated.
The demodulator is able to calculate the ratio signal on noise on each of the carriers. This is done in demodulating the carrier ~. from the values of the channel, remodulating to find the expected theoretical values and by making a difference.
Figure 9 shows the demodulation flow on a subband.
All demodulated values are stored and the demodulation information is transmitted to the controller.
All signal-to-noise ratio information is transmitted to the controller.
Control The control of the transmission solution over current is in a microcontroller that controls directly the modulator for the emission and the demodulator For the reception. The microcontroller treats different sub-bands one after the other. According to his own software configuration (embedded program loaded dynamically), it activates or not the subbands and their allocates the different data streams.
I1 does not manage the data as such, but receives data availability information and control the modulator and demodulator accordingly. Of the same way, it informs the upper layers (Coding error corrector, MAC layer, etc.) that data has been sent or received.
In addition, the controller is the sole master of applied modulations ~. each carrier and report signal on noise calculated. It is he who makes the decision to change the modulation.
~ missio ~, - odul ~ ti ~~ a The modulator is the alter ego of the demodulator. we found similar features. This is modulate data that is stored, by order of the controller. Data can be modulated by 0, 1, 2, 4, 6 or 8 depending on the quality. On each of carriers, we can forbid the modulation to respect forbidden frequencies (notion of notchs). On two pilot carriers in a symbol, one can force the modulation with known data.
Figure 10 shows the configuration of the modulation on a subband.
To limit the transmission of information, only 128 active carriers of each of the subbands are transmitted. It is downstream (especially in the reverse Fourier transform) that will be done by adding the 128 other null carriers and the construction of 256 ~ points per subband.
Multiplication The modulator requires only 6 bits of dynamics for modulation. The frequency element of each carrier is multiplied by a complex number and the dynamics is now 14 bits. The complex number used for the multiplication can break down into - an amplitude to adapt to the best electromagnetic constraints, a phase to avoid saturating the downstream calculation.
Transformed ~~ urier i ~ a ~ erse This step builds the time signal of 256 points from the 128 useful frequency elements.
sy ~ ath ~ s ~ filter:
The synthesis filter allows the translation of different data corresponding to different sub-bands in the associated frequency bands. The structure used is dual of the analysis filter used in reception the data arrive by block of 256 samples temporal, subband after subband. Following the 256 symbols, a guard interval of 32 samples temporal is added. The data of each of the sub-bands come out intertwined.
Analog part The program starts by converting the digital signal in analog.
At the output of the converter, the signal is filtered to remove the effects of the sampling frequency and avoid generating interference outside the band allocated to powerline transmission. here are the characteristics of the filter.
- Order elliptical filter 8 - Bandwidth 0-30 MHz - Stop band at -50dB
- Ripple in the bandwidth <1dB
- Bandwidth transition -> Stopband <2P2Hz Then the signal is amplified before coupling.
The amplification is fixed (gain 2).

Za Figure 11 describes the analog part of the transmission.
The invention is described in the foregoing ~ title 5 example. It is understood that the skilled person is ~ mime to realize different variants of the invention without as far out of the scope of the patent.

Claims (8)

1. Powerline data transmission method using a band of frequency, the method comprising the following steps:
- cutting said frequency band into N sub-bands, N being an integer superior or equal to two;
- implement an OFDM (Orthogonal Frequency Division) technique Multiplexing) on each of said subbands, said subbands being activated and allocated dynamically; and - perform calculation operations on each of the subbands;
characterized in that the calculations carried out on each of the subbands are independent of the calculations made on the other sub-bands;
characterized in that cutting said frequency band comprises use a filter bank adapted to cut off said N-sub frequency band.
bands;
- characterize in that performing calculation operations includes a synchronization that is performed independently starting from each said sub-bands; and characterized in that synchronization includes performing a transformed from Fourier of a first time sample to produce a first sample frequency, the calculation of a first angle of rotation for a first pilot carrier, calculating a second angle of rotation for a second pilot carrier, the deduction a third angle of rotation of a medium carrier of the first sample frequency and the establishment of the third angle of rotation for a second temporal sample before performing a Fourier transform of the second time sample, said second time sample being subsequent to the first temporal sample. 2. Method of transmitting power line data according to the claim 1, characterized in that the filters of said bank have filtered subbands overlapping. 3. Powerline data transmission method according to the claim 2, characterized in that the frequencies present in said overlaps do not are not used. 4. Power line data transmission method according to the claim 1, characterized in that the number N of sub-bands is equal to 7. 5. Power-Line Data Transmission Method According to One any of Claims 1 to 4, characterized in that the OFDM symbols have a size equal to 256 carriers. 6. A power-line data transmission method according to one of any of Claims 1 to 5, characterized in that the subbands are activated and allocated from dynamically by software means. 7. Power line data transmission method according to one of any of Claims 1 to 5, characterized in that the subbands are activated and allocated from dynamically by a material means. 8. Modulation and demodulation equipment for the implementation of the process according to any one of claims 1 to 7 comprising means for filtering, means for calculating a Fast Fourier Transform (FFT), means for Analog / Digital conversion, amplification means and means activation and allocation of subbands.